I have set up a devel/staging cluster in vmware. It isn't perfect, but it does work. I had heartbeat issues from the start because of I/O contention. What I did to resolve the problem was move one of the VMs to another VM server. The problem was that both nodes were writing to disk at the same time when checkpointing. Since we didn't have a massive disk subsystem this caused the problems. Moving the VM to another server alleviated the simultaneous writes to the same disk. The nodes haven't missed a heartbeat check since. You might not have two VMWare servers so this might not be an option. Unfortunately, the only heartbeat configuration variable I know of just determines how often the heartbeat checks are performed. Doesn't help this issue at all.

All this being said, I would never ever consider building a production cluster on VMs. Consider yourself warned :)